Help talk:Property constraints portal/Type

It’s possible that the

@D1gggg: what’s the reason for this change? --Lucas Werkmeister (WMDE) (talk) 16:41, 18 June 2017 (UTC)

Everything is stated before : "There are several possible ways..." d1g (talk) 19:54, 18 June 2017 (UTC)

But now the list is just broken English, and inconsistent with all the other constraint type help pages that also use the “It is possible…” pattern (sometimes mixed with ”The most likely case is…” or similar). I really don’t think this is an improvement. --Lucas Werkmeister (WMDE) (talk) 09:31, 19 June 2017 (UTC)

Repetitive content could be removed, I don't know how to word it.

Too much prose isn't useful in informative messages.

It should state how things are different, not how everything is the same. d1g (talk) 05:38, 20 June 2017 (UTC)

Well, if you don’t have a better suggestion, then I’m going to revert to the previous version, because the current text is definitely worse IMHO. If the “possible actions” section should be phrased differently, that should be done consistently for all constraint types. --Lucas Werkmeister (WMDE) (talk) 10:38, 20 June 2017 (UTC)

Problem with subclases

If you tak a look at for example MT6 (Q26883010) there are a lot of constraint violations, because for example the date of death (P570) are only allowed for humans, … and animals. Q26883010 is a wolf (Q18498) (what is an animal), but wolf Q18498 is not a subclass of (P279) of animal but it is a taxon (Q16521) and a subclass according parent taxon (P171). I don't see any way of fixing this constraint – do you? -- MichaelSchoenitzer (talk) 13:55, 30 July 2017 (UTC)

https://phabricator.wikimedia.org/T171933
--- Jura 15:04, 30 July 2017 (UTC)

instance or subclass relation

@Marsupium: Support for instance or subclass of (Q30208840) is not deployed to Wikidata yet, and will not be deployed for almost two more weeks (see wikitech:Deployments – Wikidata is in group1, next MediaWiki train for that is on February 7th). I’d like to wait until then to document this new relation (and then also document the “value type” page, and explain more clearly what the relation does, …). Is it okay to revert Special:Diff/623215900 for now? --Lucas Werkmeister (WMDE) (talk) 11:40, 25 January 2018 (UTC)

Sorry, I forgot that it is not yet deployed and wondered about the displayed constraint vios. :-) Reverted now. --Marsupium (talk) 11:49, 25 January 2018 (UTC)

Cool, thanks :) I’ll try to remember to ping here once the code has been deployed. --Lucas Werkmeister (WMDE) (talk) 12:49, 25 January 2018 (UTC)

 Done @Marsupium: the instance or subclass of (Q30208840) relation should be deployed now, and I’ve updated the documentation. --Lucas Werkmeister (WMDE) (talk) 14:24, 12 February 2018 (UTC)

subclassOf - instanceOf -subclassOf

I noticed that there are relations as mentioned in the title: entities that are subclass of something that is an instance of something. If I had a type constraint on an entity X with subclass of and type Y, should it be fulfilled if it is subclass of entity Z which in turn is instance of Y?

I am generally confused that subclasses of instances are allowed. Is this as intended?  – The preceding unsigned comment was added by Adrian Bielefeldt (talk • contribs) at 16:57, 25 July 2018‎ (UTC).

@Adrian Bielefeldt: I’m not sure I quite understand your specific question, I need more items and less letters :)

But in general, subclasses are always allowed, and this is very much intentional. So a constraint “instance of X” really means “instance of X, or of subclass of X, or of subclass of subclass of X, etc.”; a constraint “subclass of X” really means “subclass of X, or of subclass of X, etc.”; and “instance or subclass of X” is the union of both.

For example, a constraint with relation “instance of” and the class organization (Q43229) permits companies, associations, clubs, unions, even states, because they are all subclasses (direct or indirect) of organization (Q43229). --Lucas Werkmeister (WMDE) (talk) 16:04, 30 July 2018 (UTC)